OCTO / Voice studio · Browser edition
Write a conversation, give every speaker a voice, and hear it in English, Spanish, French, Italian, Portuguese or Hindi. It all runs in your browser. Nothing you type is uploaded.
One line per turn. Up to 4 speakers. Write each line in the language of its speaker's voice.
The first time, your browser downloads the voice model (92 MB). It is saved for next time.
Kept on this page only. Download what you want to keep.
Nothing generated yet. Your audio will appear here.
Tap a flag for the language, then choose a voice.
Voices: 41 native voices across 7 language groups, from the Kokoro 82M model (Apache-2.0), running locally through WebAssembly.
Not included here: voice cloning, Japanese and Chinese, and the larger designed voice bank. Those need the desktop app.